Update utilities/extract_tables.py

utilities/extract_tables.py

@@ -27,23 +27,36 @@
 import os
 import requests
 
-EXTRACT_TABLES_API = "https://point9-extract-text-and-table.hf.space/api/tables"  # Replace with your space URL
+# Hardcoded API URL - DO NOT CHANGE
+EXTRACT_TABLES_API = "https://point9-extract-text-and-table.hf.space/api/tables"
 
 def extract_tables_remote(state):
-    filename = state["filename"]
-    path = state["temp_files"][filename]
+    """
+    Extract tables from documents via API.
+    All file types are sent to the API (no local processing).
     
-    # Extract filename with extension from the path
-    filename_with_extension = os.path.basename(path)
+    Args:
+        state: Dict with 'filename', 'temp_files', 'start_page', 'end_page'
     
+    Returns:
+        state: Dict with 'tables' key containing list of extracted tables
+    """
+    filename = state["filename"]
+    path = state["temp_files"][filename]
+
+    if not os.path.exists(path):
+        raise RuntimeError(f"File not found: {path}")
+
     with open(path, "rb") as f:
-        files = {"file": (filename_with_extension, f, "application/pdf")}
+        files = {"file": (filename, f, "application/pdf")}
         data = {
-            "filename": filename_with_extension,  # Use filename with extension
+            "filename": filename,
             "start_page": state.get("start_page", 1),
             "end_page": state.get("end_page", 1),
         }
         headers = {"Authorization": f"Bearer {os.getenv('HUGGINGFACE_API_TOKEN')}"}
+        
+        # Call API and wait for response
         resp = requests.post(EXTRACT_TABLES_API, files=files, data=data, headers=headers)
 
     if resp.status_code != 200: